Experiment Tracking

Experiment Tracking covers the Model Training pages for running and comparing workloads, registering models, versioning datasets, and storing run outputs. Use it to keep experiment lineage and lifecycle management in one place before you manage infrastructure under Environments & Operations.

Use these pages when you need to:

• Compare training or pipeline runs and profiling metrics.
• Publish models to a registry with versions and aliases.
• Version datasets and track lineage across workflows.
• Find checkpoints, logs, and other artifacts produced by a run.

Next: start with Tasks for runs, then Models, Datasets, and Artifacts as outputs mature. For the full menu, see Model Training overview; for AI Gateway and FinOps, see Core Services.

Tasks

Tasks is where you create and monitor ML and AI workloads. A task groups related runs so you can compare executions, review metrics, and track experiment history in one place.

Use this page when you need to register a workload before profiling, check latest run status, or open a task to compare metrics.

Next: click + ADD to register a task, connect profiling via PROFILING INTEGRATION if needed, then open a task name for run details. Related outputs appear under Artifacts; trained models under Models.

Models

Models is the OptScale AI model registry—central storage for trained models, versions, aliases, and metadata used in production and experimentation.

Use this page when you need to register a model, track versions, set deployment aliases, or link models back to Tasks and Datasets.

Next: click + ADD to register a model, then open a model name to manage versions and aliases.

Datasets

Datasets manages catalog entries, versions, and lineage for training and evaluation data.

Use this page when you need to register a dataset or version, filter by how data is used or produced, or inspect lineage before a new experiment.

Next: click + ADD to register a dataset or version, then use LABEL, USED IN, and PRODUCED BY filters to narrow results.

Artifacts

Artifacts browses run outputs from training and pipeline executions—checkpoints, plots, logs, environment files, and reports. Artifacts are tied to a run and often reference external storage (for example Amazon S3).

Use this page when you need to locate outputs from a completed run or open the related run for metrics.

Next: filter by time range and TASK, then follow the Run link to related Tasks.

See also

• Environments & Operations — Shared runtimes, cloud connections, and power schedules.
• Architecture Overview — Platform structure and request flow.